Replace print() with structured logging

[maxnutz]

Problem

Warnings/info messages throughout the package are emitted via bare print() calls instead of the standard library logging module. This means:

• Output can't be filtered by severity (WARNING vs INFO all look the same, distinguished only by an ad-hoc string prefix like "WARNING: ...").
• Callers embedding this package (e.g. as part of a larger workflow) can't redirect, silence, or capture these messages via standard logging handlers.
• Tests assert on print output via capsys (tests/test_format_timestamps.py), which is brittle compared to caplog.
• No way to control verbosity (e.g. suppress INFO in production runs, or elevate to DEBUG for troubleshooting).

Current print() usage

File	Line(s)	Level (by prefix)
pypsa_validation_processing/class_definitions.py	75, 89	WARNING (format_timestamps)
pypsa_validation_processing/class_definitions.py	320	WARNING (_execute_function)
pypsa_validation_processing/class_definitions.py	669	WARNING (_get_target_unit or similar)
pypsa_validation_processing/class_definitions.py	675, 684, 686	INFO/WARNING (multi-unit parsing)
pypsa_validation_processing/class_definitions.py	723, 731, 735	INFO/WARNING (_get_network_config)
pypsa_validation_processing/statistics_functions.py	829	WARNING (BEV charger efficiency)
pypsa_validation_processing/workflow.py	30	WARNING (resolve_config_path)

This is consistent with the blueprint script sister_packages/pypsa-de/scripts/pypsa-de/export_ariadne_variables.py and sister_packages/pypsa-de/scripts/_helpers.py, which both already use logger = logging.getLogger(__name__) + logger.info/.warning/.error(...), configured centrally via logging.basicConfig(**kwargs) in _helpers.py. We should follow the same convention for consistency across the PyPSA-AT ecosystem.

Proposed implementation plan

1. Central logging setup helper — add setup_logging(level: int | str = logging.WARNING) -> None to pypsa_validation_processing/utils.py. Wraps logging.basicConfig(level=level, format=...). Called once, at the entrypoint only (workflow.py::main) — never inside statistics_functions.py, per the "statistics functions are independent, only call utils.py helpers" rule in CLAUDE.md.
2. Module-level loggers — add logger = logging.getLogger(__name__) near the top of:
   • pypsa_validation_processing/class_definitions.py
   • pypsa_validation_processing/statistics_functions.py
   • pypsa_validation_processing/workflow.py
3. Replace print() calls with the matching logger.warning(...) / logger.info(...) call, dropping the manual "WARNING: " / "INFO: " string prefixes (the log level now carries that information; formatters can add it back for console output if desired).
4. Wire up setup_logging() in workflow.py::main() before Network_Processor is constructed, defaulting to WARNING. Do not add a new log_level config key to configs/*.yaml as part of this change (config files are off-limits per CLAUDE.md unless explicitly requested) — default level is fine for a first pass; a follow-up issue can add config-driven verbosity if desired.
5. Update tests — tests/test_format_timestamps.py::test_format_timestamps_sets_nat_on_localization_failure currently uses capsys; switch it to caplog and assert on the log record's level/message instead of stdout text. Add/extend unit tests for the other converted call sites where they're already covered.
6. Docs — add a short "Logging" note to docs/contributing.md (or README.md) describing the logger = logging.getLogger(__name__) convention for new statistics functions/helpers, so future contributions don't reintroduce print().

Out of scope

• Anything under sister_packages/ (read-only, per CLAUDE.md).
• Adding a log_level config option to configs/*.yaml (separate follow-up if wanted).
• Structured/JSON logging, log file rotation, or third-party logging libraries (e.g. loguru) — plain stdlib logging matches the PyPSA-DE blueprint and needs no new dependency.

Acceptance criteria

• No print( calls remain in pypsa_validation_processing/*.py (excluding sister_packages/).
• Each affected module has its own logger = logging.getLogger(__name__).
• workflow.py::main() configures logging once via utils.setup_logging().
• pixi run test passes, including the updated caplog-based test.
• pixi run workflow_test runs successfully with log output visible at WARNING level by default.